Notification Server killing the SQL server!

How backed up are your event queues (EvtQFast, EvtQueue,etc.)?  It sounds like you may be trying to squeeze too much out of the server.

You may want to consider moving SQL off-box.  Asset is pretty SQL intensive.

have you implemented the /3GB switch in the boot.ini? Otherwise, you're limited what memory you do have available. Have you limited how much memory SQL can use? In line with what dfrancis said, check your Altiris log files to see if  you are getting any errors,

I checked the event queues, and new files are created each 3-5 minutes, and this is inventory files, and loging/logoff events (and others events). Nothing to kill the SQL server here?!

And about the /3GB, nope, it is missing from the boot.ini file. I will configure that tonight...

Actually the sqlservr.exe process is taking 1.8 GB if memory, which our DBA consider normal (SQL Server 2000 Std / Windows Server 2003 Std).

The Altiris event log is not showing anything wrong about the Notification Server -> SQL...

What's wrong with our Notification Server???

Thanks.

Claude Lachapelle
Systems Administrator, La Coop fédérée

I re-enabled "Enable Notification Server Processing", and all .tmp file in the queues have been processed (approx. 600 files since this morning).

The disk queue lenght raised juste a little, cpu and page/sec reached 70/80% but only for a few sec.

Now, from the SQL Profiler I could see more and more Altiris Notification Server batch processed... and guess what? Disk queue lenght is now back to 30-40... steady!!!

Here one of those Altiris Notification Server batch:

exec spResourceUpdateSummary '18f8e9a1-639b-422d-9ee6-ee41c9debda1', null, '57beb323-f925-4689-872f-1bf3aa3f7632', null

As soon as I disabled the "Enable Notification Server Processing", .tmp are created into queues directory, and after a few minutes, disk queue lenght is getting back to normal (2-3 max.). No more Altiris Notification Server process into SQL Profiler...

Do we have a problem with our SQL server disk systems? Everything is on the same disk (RAID-5), logs, data and os (where program files resides too!). I suspect this is the problem... to much I/O at the same time. I will try to seperate the os and logs from the sql database...

Thanks.

Claude Lachapelle
Systems Administrator, La Coop fédérée

SQL log files & pagefile.sys are now on a separate disk.

All Windows patches applied (including SP2).

Still having high disk queue lenght problem where SQL data files are.

Now running Index Tuning Wizard to find out if we could create new indexes to accelerate SQL read performance (80% read / 20% write).

And the server is peaking always when the Altiris Notification Server is running a lot of these procedures:

 exec spResourceUpdateSummary '50663863-8c9f-4663-865f-6c524c4a46bb', null, '57beb323-f925-4689-872f-1bf3aa3f7632', null

Any idea?

Thanks,

Claude Lachapelle
Systems Administrator, La Coop fédérée

Finally found the problem... a scheduled task was running every 10 minutes to get users from the Active Directory, but it takes 12 minutes to complete!!!